Framing the Future First: Using Query Theory to Explore Claiming - - PowerPoint PPT Presentation

Melissa Knoll* 10/4/2012

Financial Literacy Seminar Series George Washington University

*Joint work with Kirstin Appelt, Eric Johnson, and Jon Westfall Center for Decision Sciences, Columbia University **The opinions expressed are those of the authors and do not represent the views of the Social Security Administration.

Framing the Future First:

Using “Query Theory” to Explore Claiming Preferences

Retiring in America

 The Social Security Administration (SSA) is neutral with regard to when

individuals should claim retirement benefits.

 40-50% of Americans claim benefits as soon as they are eligible

(Muldoon & Kopcke, 2008; Song & Manchester, 2007).

 Many Americans are financially underprepared for a long retirement

(EBRI, 2010; Thaler & Benartzi, 2004).

 SS retirement benefits are a primary source of income for over 50% of

• lder Americans (SSA, 2010).

Claiming Retirement Benefits

 Benefits can be claimed at any time after age 62  The longer claiming is delayed, the larger the monthly benefit  Claiming is an intertemporal choice

 Claiming is a choice between smaller benefit sooner and larger benefit later

Query Theory (Johnson et al., 2007; Weber et al., 2007)

 Memory-based model of preference construction

 People construct preferences “on the fly”  They construct preferences by asking themselves about pros and cons  Decisions often have natural reference point/default

 Initial thoughts biased in favor of reference point  Subsequent thoughts influenced by this bias (suppressed)  Reference point predicts order and balance of thoughts, which

predict choice

Query Theory & Claiming

 Early claiming is natural reference point for many Americans

 According to QT…

 people will first query their memory for arguments in favor of claiming early

(e.g., “declining health” or “prefer early retirement”),

 then they will query their memory for arguments in favor of delaying claiming

(e.g., “I would like to work as long as I can” or “I want my full benefits”).

QT, Framing, and Claiming

 Change reference point (early claiming) by changing the frame  According to QT…

change reference point  change query order change preferences

Overview of Studies

 Study 1: Explore claiming decision  Study 2: Display change: Mild Framing  Study 3: Display change: Extreme Framing  Study 4: Process change (1): QT  Study 5: Process change (2): A QT checklist

Study 1: Exploratory Study

 Many participants will prefer to claim early (replicate survey

data—spikes at 62 and FRA)

 Early claiming will be reference point for many participants  Prominence of thoughts in favor of early claiming will predict

preference for early claiming

Many participants prefer to claim early

0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 62 63 64 65 66 67 68 69 70 Percent of Participants Preferred Claiming Age Not Yet Eligible Eligible

• 47% of participants preferred to claim benefits early (i.e., before FRA)
• 33% of participants preferred to claim benefits as early as possible
• Not yet eligible—36% preferred to claim early and 26% preferred to claim as early as possible
• Eligible—61% preferred to claim early and 42% preferred to claim as early as possible

Early claiming is a reference point for many participants

0% 10% 20% 30% 40% 50% 60% 70% Early claiming Full claiming Delayed claiming Percent of Participants Reference Point Not Yet Eligible Eligible

• Which claiming age and benefit amount combination did they think about the most?
• More participants used early claiming as a reference point (51%) than full (21%) or delayed claiming (28%)
• Reference point and preferred claiming age were highly correlated, r(1080) = .78, p < .001.

Prominence of thoughts in favor of early claiming predicts preferences for early claiming Prominence of Early-Claiming Thoughts Preferred Claiming Age

β = -.68, p < .001

Exploratory Study—Main Findings

 Replicate retirement spikes  Many participants adopt early claiming as reference point  Prominence of early-claiming thoughts predicts preference for

early claiming

Study 2: Mild Frame Change

 Many people use early claiming as reference point  What happens to claiming preferences if we change the information

display?

Participants in the shifted axis condition do not adopt later claiming as a reference point

0% 10% 20% 30% 40% 50% Early-as-possible claiming Early claiming Full claiming Delayed claiming Percent of Participants Standard Axis Shifted Axis

p > .5 *Note: We added an “early-as-possible claiming” option

Participants in the shifted axis condition do not prefer to claim later

65.20 65.29 62 63 64 65 66 67 68 69 70 Standard Axis Shifted Axis Mean Preferred Claiming Age

p > .2

Study 3: Extreme Frame Change

Participants seeing the extreme graph do not prefer to claim later

Study 4: QT Process Change

 Many people use early claiming as reference point  What happens to claiming preferences if we change how

people approach the decision?

Ask participants to “frame the future first”

 Natural order: Participants list early-claiming thoughts

first and later-claiming thoughts second

 Unnatural order: Participants list later-claiming thoughts

first and early-claiming thoughts second

Study 4: QT Process Change

 Participants in unnatural order will…

 adopt later claiming as reference point  have less prominent thoughts about early claiming  prefer to claim later

Participants in the unnatural order adopt later claiming as a reference point

0% 10% 20% 30% 40% 50% 60% 70% Early-as- possible claiming Early claiming Full claiming Delayed claiming Percent of Participants Reference Point Natural Order Unnatural Order

t(214) = -3.32, p = .001, Cohen’s d = .45

Participants in the unnatural order have less prominent thoughts in favor of early claiming

• 0.80
• 0.60
• 0.40
• 0.20

0.00 0.20 0.40 0.60 0.80 Natural Order Unnatural Order Mean Prominence of Early-Claiming Thoughts

t(283.1) = 25.01, p < .001, Cohen’s d = 2.77

Participants in the unnatural order prefer to claim later (approx. 9 months)

62 63 64 65 66 67 68 69 70 Natural Order Unnatural Order Mean Preferred Claiming Age

t(319.9) = -2.47, p = .01, Cohen’s d = .28

64.81 65.59

Study 5: QT Checklist

Checklist order significantly affected claiming age

62 63 64 65 66 67 68 69 70 Natural Order Unnatural Order 65.91

Mean preferred claiming age

65.28

Effectiveness of Interventions

• 20
• 15
• 10
• 5

5 10 15

Brown, Kapetyn & Mitchell (2011)

Months relative to control

Summary

 The claiming decision is extremely important, but:  It seems to be constructed  44% say they first think about it within a year of retiring (EBRI, 2008)  Suggests it can be changed  The reference point is often early claiming  Changing the frame can change the reference point  Query Theory says:

*change reference point, change query order, change preferences*

 Understanding the process can help design interventions.  Changing displays: Not yet  Changing the order of consideration (“frame the future first”): Yes!

Thank you! Questions? Comments?